git.sesse.net Git - vlc/blob - modules/codec/subtitles/subsass.c

   1 /*****************************************************************************
   2  * subsass.c : ASS/SSA subtitles decoder
   3  *****************************************************************************
   4  * Copyright (C) 2000-2006 the VideoLAN team
   5  * $Id$
   6  *
   7  * Authors: Gildas Bazin <gbazin@videolan.org>
   8  *          Samuel Hocevar <sam@zoy.org>
   9  *          Derk-Jan Hartman <hartman at videolan dot org>
  10  *
  11  * This program is free software; you can redistribute it and/or modify
  12  * it under the terms of the GNU General Public License as published by
  13  * the Free Software Foundation; either version 2 of the License, or
  14  * (at your option) any later version.
  15  *
  16  * This program is distributed in the hope that it will be useful,
  17  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  18  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19  * GNU General Public License for more details.
  20  *
  21  * You should have received a copy of the GNU General Public License
  22  * along with this program; if not, write to the Free Software
  23  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
  24  *****************************************************************************/
  25 #ifdef HAVE_CONFIG_H
  26 # include "config.h"
  27 #endif
  28
  29 #include "subsdec.h"
  30
  31
  32 void ParseSSAString( decoder_t *p_dec,
  33                      char *psz_subtitle,
  34                      subpicture_t *p_spu_in )
  35 {
  36     /* We expect MKV formatted SSA:
  37      * ReadOrder, Layer, Style, CharacterName, MarginL, MarginR,
  38      * MarginV, Effect, Text */
  39     decoder_sys_t   *p_sys = p_dec->p_sys;
  40     subpicture_t    *p_spu = p_spu_in;
  41     ssa_style_t     *p_style = NULL;
  42     char            *psz_new_subtitle = NULL;
  43     char            *psz_buffer_sub = NULL;
  44     char            *psz_style = NULL;
  45     char            *psz_style_start = NULL;
  46     char            *psz_style_end = NULL;
  47     int             i_text = 0, i_comma = 0, i_strlen = 0, i;
  48     int             i_margin_l = 0, i_margin_r = 0, i_margin_v = 0;
  49
  50     psz_buffer_sub = psz_subtitle;
  51
  52     p_spu->p_region->psz_html = NULL;
  53
  54     i_comma = 0;
  55     while( i_comma < 8 && *psz_buffer_sub != '\0' )
  56     {
  57         if( *psz_buffer_sub == ',' )
  58         {
  59             i_comma++;
  60             if( i_comma == 2 )
  61                 psz_style_start = &psz_buffer_sub[1];
  62             else if( i_comma == 3 )
  63                 psz_style_end = &psz_buffer_sub[0];
  64             else if( i_comma == 4 )
  65                 i_margin_l = (int)strtol( &psz_buffer_sub[1], NULL, 10 );
  66             else if( i_comma == 5 )
  67                 i_margin_r = (int)strtol( &psz_buffer_sub[1], NULL, 10 );
  68             else if( i_comma == 6 )
  69                 i_margin_v = (int)strtol( &psz_buffer_sub[1], NULL, 10 );
  70         }
  71         psz_buffer_sub++;
  72     }
  73
  74     if( *psz_buffer_sub == '\0' && i_comma == 8 )
  75     {
  76         msg_Dbg( p_dec, "couldn't find all fields in this SSA line" );
  77         return;
  78     }
  79
  80     psz_new_subtitle = malloc( strlen( psz_buffer_sub ) + 1);
  81     i_text = 0;
  82     while( psz_buffer_sub[0] != '\0' )
  83     {
  84         if( psz_buffer_sub[0] == '\\' && psz_buffer_sub[1] == 'n' )
  85         {
  86             psz_new_subtitle[i_text] = ' ';
  87             i_text++;
  88             psz_buffer_sub += 2;
  89         }
  90         else if( psz_buffer_sub[0] == '\\' && psz_buffer_sub[1] == 'N' )
  91         {
  92             psz_new_subtitle[i_text] = '\n';
  93             i_text++;
  94             psz_buffer_sub += 2;
  95         }
  96         else if( psz_buffer_sub[0] == '{' &&
  97                  psz_buffer_sub[1] == '\\' )
  98         {
  99             /* SSA control code */
 100             while( psz_buffer_sub[0] != '\0' &&
 101                    psz_buffer_sub[0] != '}' )
 102             {
 103                 psz_buffer_sub++;
 104             }
 105             psz_buffer_sub++;
 106         }
 107         else
 108         {
 109             psz_new_subtitle[i_text] = psz_buffer_sub[0];
 110             i_text++;
 111             psz_buffer_sub++;
 112         }
 113     }
 114     psz_new_subtitle[i_text] = '\0';
 115
 116     i_strlen = __MAX( psz_style_end - psz_style_start, 0);
 117     psz_style = strndup( psz_style_start, i_strlen );
 118
 119     for( i = 0; i < p_sys->i_ssa_styles; i++ )
 120     {
 121         if( !strcmp( p_sys->pp_ssa_styles[i]->psz_stylename, psz_style ) )
 122             p_style = p_sys->pp_ssa_styles[i];
 123     }
 124     if( psz_style ) free( psz_style );
 125
 126     p_spu->p_region->psz_text = psz_new_subtitle;
 127     if( p_style == NULL )
 128     {
 129         p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 130         p_spu->i_x = p_sys->i_align ? 20 : 0;
 131         p_spu->i_y = 10;
 132     }
 133     else
 134     {
 135         msg_Dbg( p_dec, "style is: %s", p_style->psz_stylename);
 136         p_spu->p_region->p_style = &p_style->font_style;
 137         p_spu->p_region->i_align = p_style->i_align;
 138         if( p_style->i_align & SUBPICTURE_ALIGN_LEFT )
 139         {
 140             p_spu->i_x = (i_margin_l) ? i_margin_l : p_style->i_margin_h;
 141         }
 142         else if( p_style->i_align & SUBPICTURE_ALIGN_RIGHT )
 143         {
 144             p_spu->i_x = (i_margin_r) ? i_margin_r : p_style->i_margin_h;
 145         }
 146         p_spu->i_y = (i_margin_v) ? i_margin_v : p_style->i_margin_v;
 147     }
 148 }
 149
 150 /*****************************************************************************
 151  * ParseColor: SSA stores color in BBGGRR, in ASS it uses AABBGGRR
 152  * The string value in the string can be a pure integer, or hexadecimal &HBBGGRR
 153  *****************************************************************************/
 154 static void ParseColor( decoder_t *p_dec,
 155                         char *psz_color,
 156                         int *pi_color,
 157                         int *pi_alpha )
 158 {
 159     int i_color = 0;
 160     if( !strncasecmp( psz_color, "&H", 2 ) )
 161     {
 162         /* textual HEX representation */
 163         i_color = (int) strtol( psz_color+2, NULL, 16 );
 164     }
 165     else i_color = (int) strtol( psz_color, NULL, 0 );
 166
 167     *pi_color = 0;
 168     *pi_color |= ( ( i_color & 0x000000FF ) << 16 ); /* Red */
 169     *pi_color |= ( ( i_color & 0x0000FF00 ) );       /* Green */
 170     *pi_color |= ( ( i_color & 0x00FF0000 ) >> 16 ); /* Blue */
 171
 172     if( pi_alpha != NULL )
 173         *pi_alpha = ( i_color & 0xFF000000 ) >> 24;
 174 }
 175
 176 /*****************************************************************************
 177  * ParseSSAHeader: Retrieve global formatting information etc
 178  *****************************************************************************/
 179 void ParseSSAHeader( decoder_t *p_dec )
 180 {
 181     decoder_sys_t *p_sys = p_dec->p_sys;
 182     char *psz_parser = NULL;
 183     char *psz_header = malloc( p_dec->fmt_in.i_extra+1 );
 184     int i_section_type = 1;
 185
 186     memcpy( psz_header, p_dec->fmt_in.p_extra, p_dec->fmt_in.i_extra );
 187     psz_header[ p_dec->fmt_in.i_extra] = '\0';
 188
 189     /* Handle [Script Info] section */
 190     psz_parser = strcasestr( psz_header, "[Script Info]" );
 191     if( psz_parser == NULL ) goto eof;
 192
 193     psz_parser = GotoNextLine( psz_parser );
 194
 195     while( psz_parser[0] != '\0' )
 196     {
 197         int temp;
 198         char buffer_text[MAX_LINE + 1];
 199
 200         if( psz_parser[0] == '!' || psz_parser[0] == ';' ) /* comment */;
 201         else if( sscanf( psz_parser, "PlayResX: %d", &temp ) == 1 )
 202             p_sys->i_original_width = ( temp > 0 ) ? temp : -1;
 203         else if( sscanf( psz_parser, "PlayResY: %d", &temp ) == 1 )
 204             p_sys->i_original_height = ( temp > 0 ) ? temp : -1;
 205         else if( sscanf( psz_parser, "Script Type: %8192s", buffer_text ) == 1 )
 206         {
 207             if( !strcasecmp( buffer_text, "V4.00+" ) ) p_sys->b_ass = VLC_TRUE;
 208         }
 209         else if( !strncasecmp( psz_parser, "[V4 Styles]", 11 ) )
 210             i_section_type = 1;
 211         else if( !strncasecmp( psz_parser, "[V4+ Styles]", 12) )
 212         {
 213             i_section_type = 2;
 214             p_sys->b_ass = VLC_TRUE;
 215         }
 216         else if( !strncasecmp( psz_parser, "[Events]", 8 ) )
 217             i_section_type = 4;
 218         else if( !strncasecmp( psz_parser, "Style:", 6 ) )
 219         {
 220             int i_font_size, i_bold, i_italic, i_border, i_outline, i_shadow,
 221                 i_underline, i_strikeout, i_scale_x, i_scale_y, i_spacing,
 222                 i_align, i_margin_l, i_margin_r, i_margin_v;
 223
 224             char psz_temp_stylename[MAX_LINE+1];
 225             char psz_temp_fontname[MAX_LINE+1];
 226             char psz_temp_color1[MAX_LINE+1];
 227             char psz_temp_color2[MAX_LINE+1];
 228             char psz_temp_color3[MAX_LINE+1];
 229             char psz_temp_color4[MAX_LINE+1];
 230
 231             if( i_section_type == 1 ) /* V4 */
 232             {
 233                 if( sscanf( psz_parser, "Style: %8192[^,],%8192[^,],%d,%8192[^,],%8192[^,],%8192[^,],%8192[^,],%d,%d,%d,%d,%d,%d,%d,%d,%d%*[^\r\n]",
 234                     psz_temp_stylename, psz_temp_fontname, &i_font_size,
 235                     psz_temp_color1, psz_temp_color2, psz_temp_color3,
 236                     psz_temp_color4, &i_bold, &i_italic,
 237                     &i_border, &i_outline, &i_shadow, &i_align, &i_margin_l,
 238                     &i_margin_r, &i_margin_v ) == 16 )
 239                 {
 240                     ssa_style_t *p_style = malloc( sizeof(ssa_style_t) );
 241
 242                     p_style->psz_stylename = strdup( psz_temp_stylename );
 243                     p_style->font_style.psz_fontname = strdup( psz_temp_fontname );
 244                     p_style->font_style.i_font_size = i_font_size;
 245
 246                     ParseColor( p_dec, psz_temp_color1, &p_style->font_style.i_font_color, NULL );
 247                     ParseColor( p_dec, psz_temp_color4, &p_style->font_style.i_shadow_color, NULL );
 248                     p_style->font_style.i_outline_color = p_style->font_style.i_shadow_color;
 249                     p_style->font_style.i_font_alpha = p_style->font_style.i_outline_alpha
 250                                                      = p_style->font_style.i_shadow_alpha = 0x00;
 251                     p_style->font_style.i_style_flags = 0;
 252                     if( i_bold ) p_style->font_style.i_style_flags |= STYLE_BOLD;
 253                     if( i_italic ) p_style->font_style.i_style_flags |= STYLE_ITALIC;
 254
 255                     if( i_border == 1 )
 256                         p_style->font_style.i_style_flags |= (STYLE_ITALIC | STYLE_OUTLINE);
 257                     else if( i_border == 3 )
 258                     {
 259                         p_style->font_style.i_style_flags |= STYLE_BACKGROUND;
 260                         p_style->font_style.i_background_color = p_style->font_style.i_shadow_color;
 261                         p_style->font_style.i_background_alpha = p_style->font_style.i_shadow_alpha;
 262                     }
 263                     p_style->font_style.i_shadow_width = i_shadow;
 264                     p_style->font_style.i_outline_width = i_outline;
 265
 266                     p_style->i_align = 0;
 267                     if( i_align == 1 || i_align == 5 || i_align == 9 )
 268                         p_style->i_align |= SUBPICTURE_ALIGN_LEFT;
 269                     if( i_align == 3 || i_align == 7 || i_align == 11 )
 270                         p_style->i_align |= SUBPICTURE_ALIGN_RIGHT;
 271                     if( i_align < 4 )
 272                         p_style->i_align |= SUBPICTURE_ALIGN_BOTTOM;
 273                     else if( i_align < 8 )
 274                         p_style->i_align |= SUBPICTURE_ALIGN_TOP;
 275
 276                     p_style->i_margin_h = ( p_style->i_align & SUBPICTURE_ALIGN_RIGHT ) ?
 277                                                         i_margin_r : i_margin_l;
 278                     p_style->i_margin_v = i_margin_v;
 279                     p_style->i_margin_percent_h = 0;
 280                     p_style->i_margin_percent_v = 0;
 281
 282                     p_style->font_style.i_karaoke_background_color = 0xffffff;
 283                     p_style->font_style.i_karaoke_background_alpha = 0xff;
 284
 285                     TAB_APPEND( p_sys->i_ssa_styles, p_sys->pp_ssa_styles, p_style );
 286                 }
 287                 else msg_Warn( p_dec, "SSA v4 styleline parsing failed" );
 288             }
 289             else if( i_section_type == 2 ) /* V4+ */
 290             {
 291                 /* Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour,
 292                    Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline,
 293                    Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
 294                 */
 295                 if( sscanf( psz_parser, "Style: %8192[^,],%8192[^,],%d,%8192[^,],%8192[^,],%8192[^,],%8192[^,],%d,%d,%d,%d,%d,%d,%d,%*f,%d,%d,%d,%d,%d,%d,%d%*[^\r\n]",
 296                     psz_temp_stylename, psz_temp_fontname, &i_font_size,
 297                     psz_temp_color1, psz_temp_color2, psz_temp_color3, psz_temp_color4, &i_bold, &i_italic,
 298                     &i_underline, &i_strikeout, &i_scale_x, &i_scale_y, &i_spacing, &i_border, &i_outline,
 299                     &i_shadow, &i_align, &i_margin_l, &i_margin_r, &i_margin_v ) == 21 )
 300                 {
 301                     ssa_style_t *p_style = malloc( sizeof(ssa_style_t) );
 302
 303                     p_style->psz_stylename = strdup( psz_temp_stylename );
 304                     p_style->font_style.psz_fontname = strdup( psz_temp_fontname );
 305                     p_style->font_style.i_font_size = i_font_size;
 306                     msg_Dbg( p_dec, psz_temp_color1 );
 307                     ParseColor( p_dec, psz_temp_color1, &p_style->font_style.i_font_color,
 308                                 &p_style->font_style.i_font_alpha );
 309                     ParseColor( p_dec, psz_temp_color3, &p_style->font_style.i_outline_color,
 310                                 &p_style->font_style.i_outline_alpha );
 311                     ParseColor( p_dec, psz_temp_color4, &p_style->font_style.i_shadow_color,
 312                                 &p_style->font_style.i_shadow_alpha );
 313
 314                     p_style->font_style.i_style_flags = 0;
 315                     if( i_bold ) p_style->font_style.i_style_flags |= STYLE_BOLD;
 316                     if( i_italic ) p_style->font_style.i_style_flags |= STYLE_ITALIC;
 317                     if( i_underline ) p_style->font_style.i_style_flags |= STYLE_UNDERLINE;
 318                     if( i_strikeout ) p_style->font_style.i_style_flags |= STYLE_STRIKEOUT;
 319                     if( i_border == 1 ) p_style->font_style.i_style_flags |= (STYLE_ITALIC | STYLE_OUTLINE);
 320                     else if( i_border == 3 )
 321                     {
 322                         p_style->font_style.i_style_flags |= STYLE_BACKGROUND;
 323                         p_style->font_style.i_background_color = p_style->font_style.i_shadow_color;
 324                         p_style->font_style.i_background_alpha = p_style->font_style.i_shadow_alpha;
 325                     }
 326                     p_style->font_style.i_shadow_width  = ( i_border == 1 ) ? i_shadow : 0;
 327                     p_style->font_style.i_outline_width = ( i_border == 1 ) ? i_outline : 0;
 328                     p_style->font_style.i_spacing = i_spacing;
 329                     //p_style->font_style.f_angle = f_angle;
 330
 331                     p_style->i_align = 0;
 332                     if( i_align == 0x1 || i_align == 0x4 || i_align == 0x7 )
 333                         p_style->i_align |= SUBPICTURE_ALIGN_LEFT;
 334                     if( i_align == 0x3 || i_align == 0x6 || i_align == 0x9 )
 335                         p_style->i_align |= SUBPICTURE_ALIGN_RIGHT;
 336                     if( i_align == 0x7 || i_align == 0x8 || i_align == 0x9 )
 337                         p_style->i_align |= SUBPICTURE_ALIGN_TOP;
 338                     if( i_align == 0x1 || i_align == 0x2 || i_align == 0x3 )
 339                         p_style->i_align |= SUBPICTURE_ALIGN_BOTTOM;
 340                     p_style->i_margin_h = ( p_style->i_align & SUBPICTURE_ALIGN_RIGHT ) ?
 341                                             i_margin_r : i_margin_l;
 342                     p_style->i_margin_v = i_margin_v;
 343                     p_style->i_margin_percent_h = 0;
 344                     p_style->i_margin_percent_v = 0;
 345
 346                     p_style->font_style.i_karaoke_background_color = 0xffffff;
 347                     p_style->font_style.i_karaoke_background_alpha = 0xff;
 348
 349                     /*TODO: Ignored: angle i_scale_x|y (fontscaling), i_encoding */
 350                     TAB_APPEND( p_sys->i_ssa_styles, p_sys->pp_ssa_styles, p_style );
 351                 }
 352                 else msg_Dbg( p_dec, "SSA V4+ styleline parsing failed" );
 353             }
 354         }
 355         psz_parser = GotoNextLine( psz_parser );
 356     }
 357
 358 eof:
 359     if( psz_header ) free( psz_header );
 360     return;
 361 }
 362
 363